rogelio e. cardona-rivera, ph.d. - gfi: a formal approach ......rogelio e. cardona-rivera1; 2, jos e...

GFI: A Formal Approach to Narrative Designand Game Research

Rogelio E. Cardona-Rivera1,2, Jose P. Zagal2, and Michael S. Debus3

1 School of Computing2 The Entertainment Arts and Engineering Program

University of Utah, Salt Lake City, UT, [email protected], [email protected]

3 IT University of Copenhagen, Copenhagen, [email protected]

Abstract. We present the GFI framework (standing for Goals, Feed-back, and Interpretation), inductively developed to address the MDAframework’s shortcomings as a lens and tool for modeling games. GFIstands parallel to MDA as a formal approach that offers bridging the gapbetween narrative design, game development, story analysis, and gameresearch. Through GFI, we analyze foundational narrative design prob-lems and systematically peek through the game design space, in orderto evidence its analytical and generative traction. We believe these dis-cussions signal GFI’s potential to elucidate the narrative design process,making it easier for researchers and practitioners to decompose, study,and design a broad class of game artifacts.

Keywords: goals, feedback, interpretation, narrative design, games

1 Introduction

Narrative designers aim to better-integrate narrative and gameplay [26, 44], butthe role remains contested. They are not necessarily “writers” or “authors,” butare responsible for aspects of each; to different degrees depending on genre [28].Is environmental storytelling [33] part of the job? What about writing cut-scenescreenplays? Should they contribute to game design documents? [30] How shouldthey manage the relationship between a game’s ludic and narrative elements? [63,77, 73] What knowledge should they possess? [74]

Thus, the scope and extent of a narrative designer’s role remains unclear.This knowledge gap is critical: we cannot build a cohesive body of narrativedesign knowledge without knowing what its scope should be. As a consequence,we do not know how to best support narrative design. This paper articulatesa framework of concepts we argue are central to this practice. To us, a nar-rative designer is concerned with elements beyond Mechanics, Dynamics, andAesthetics, or MDA [47]—itself the dominant formal approach to game designand game research. We argue they also must be concerned with Goals, Feedback,and Interpretation, or GFI—our proposed formal approach to narrative design.

2 R. E. Cardona-Rivera et al.

2 Related Work

GFI is most similar (and complementary) to the MDA framework. MDA decom-poses a game into three design components: (1) Mechanics, “the components ofthe game at the level of data representation and algorithms” that reflect the rules,(2) Dynamics, “the run-time behavior of the mechanics acting on player inputsand each others’ outputs over time” that reflects the system, and (3) Aesthetics,“the desirable emotional responses evoked in the player, when she interacts withthe game system” that reflect the “fun.”

Current works that conceptually describe interactive narrative games are of-ten interactive narrative-specific accounts of MDA. Aarseth’s [2] Narrative The-ory of Games is an (interactive narrative-specific) model of the Mechanics: it de-scribes components of games across a narrative-theoretic ontology that includesthe story world (space), events, and entities (objects and agents). Koenitz’s [54]SPP framework is a model of the Dynamics: it describes the behavior of gamesrelative to the player’s interaction with a System (of potential stories), whichtriggers a Process (of interactive story-refinement) that concludes as a Product(the effected narrative). Punday [71] characterizes Involvement, Interruption,and Inevitability (III) as central to the emotional force that an interactive nar-rative game carries; III is a model of the Aesthetics. Because they focus ondifferent components of MDA, these works by Aarseth, Koenitz, and Pundayare complementary; between themselves and to our work. GFI fills in vocabu-lary necessary to describe phenomena relevant to all—but crucial to interactivenarrative—games, whose elements we posit are central to narrative design.

Game design schemas (and MDA in particular) conflict with narrative designones [36, for example]: “...by separating narrative and game modes as distinctphenomena, integrated use has been thwarted...a siloing of game developmentroles, and ultimately, functions within a game” [26, p.33]. Designing games likethis leads to the narrative wrapper [26]: narrative that “wraps around the game-play to make it transportable and attractive,” ultimately “unattached and dis-posable” [23]. We briefly discuss how MDA falls short of describing elementsrelevant to narrative design, propose GFI as a framework for describing thoseelements, and evidence GFI’s analytical utility through several case examples.

3 GFI

GFI complements MDA. In MDA, Mechanics describe how players can act,not why they would want to. In GFI, Goals (§3.1) models player motivationsand intentions, which are key for (ludic [38] and) narrative engagement [9, 65].Further, MDA’s Dynamics describe what system behavior results from playerinputs, not how to elicit inputs that are supported by the system. These inputsare determined by the tight-coupling of players to the perceivable features oftheir environments [61]. In GFI, Feedback (§3.2) models these features, criticalto structuring the player’s activity. Finally, MDA’s Aesthetics describe whatplayers feel, not what leads to it. Evoked emotions arise from what players thinkabout, which presupposes an Interpretation (§3.3) of their experience [21].

GFI: A Formal Approach to Narrative Design and Game Research 3

3.1 Goals

Goals are conditions players are expected to meet to succeed at a game. They arewidely thought of as structurally key to games [79], and have two broad senses(player-defined [12] ones are out of scope). Ludological goals are codified andrecognized in-game [25]. Narrative goals are player-interpretations of ludologicalones [19]. We discuss their ludological sense next; the narrative sense is in §3.3.

There are two kinds of ludological goals. All games have ultimate goals thatdetermine their end conditions [84]. There are at least three (Table 1): Win agame (of Chess), Finish a game (of Super Mario Bros. [27, SMB ]), or Prolongthe act of playing (e.g. by surviving in DayZ [50]).

Table 1. Ultimate goals: conditions that determine a game’s end [84].

Ultimate Description (“Games with this ultimate goal...”)

Win Effect an evaluation when a predefined state is reached.Finish Effect no evaluation when a predefined state is reached.Prolong Conclude against the designer or player’s intent.

Achieving a game’s ultimate goal requires satisfying the proximate [78] orimperative [25] goal, that it necessarily decomposes into, whose accomplishmententails the ultimate’s. These imperative goals more-concretely require the playerto effect a particular game state of affairs codified in the game itself [25]. Thereare at least 10 types (Table 2): Choose, Configure, Create, Find, Obtain, Opti-mize, Reach, Remove, Solve, and Synchronize. An imperative links game elementssuch as space, time, and entities [24]. Each one has a logical dual: its prevention.

Table 2. Imperative goals: conditions necessary to achieve a game’s ultimate goal [25].

Imperative Description (“This imperative requires players to...”)

Choose Select one element from a finite set of elements.Configure Manipulate elements such that they are in a “correct” state.Create Bring an element into existence that was not before.Find Locate a particular element.Obtain Bring a particular element under control.Optimize Accumulate a requested amount of a particular element.Reach Navigate to a particular location.Remove Eliminate an element from existence that existed before.Solve Select one “correct” element from an infinite set of elements.Synchronize Bring one or more elements into spatial/temporal unity.


Imperatives may infinitely decompose into more-specific others, creating aLudological Goal Hierarchy [19]. The hierarchy’s base maps onto a moment ingameplay. Fig. 1 illustrates this idea: to Finish SMB, one must Remove (theagent depicted as) Bowser. To do so, one must Reach the axe. To do so, (inFig. 1’s state) one might want to Prevent (spatiotemporally) Synchronizing withthe fireball. To do so, one might need to Reach the platform. And so on.

Fig. 1. Finish SMB requires Remove-Bowser via the more-specific imperativeReach-Axe, closer to the needed gameplay.

Fig. 2. The Denotational Feedback at theend of SMB elicits one interpretation ofits ultimate goal: “Save the princess.”

Analytically identifying a game’s full hierarchy is challenging, since it mustencompass all ludological goals a player may face in all possible playthroughs.However, ludological goals (and their hierarchy) are under the direct control ofa designer who specifies what conditions “count” to satisfy the ludological goals.Next, we discuss how careful construction of feedback is required such that goalsare communicated to the player in order to motivate their activity.

3.2 Feedback

Feedback is the designed multi-modal stimuli intended to convey perceptualinformation about the game’s structural elements: its underlying ends (goals)and the available means to achieve them (mechanics). This may include thingssuch as graphics, music, sound, text, and more.

Crafting feedback is arguably a narrative designer’s most critical responsi-bility, and is what makes the practice of narrative design relevant to all but themost abstract games, including ones that do not necessarily place a primacyon narrative. (Narrative designer) Dansky [23] argues that people only think ofexplicitly denoted narrative elements as the game’s narrative. However:

There’s also implicit narrative built into every game through the choice ofsetting, items, character design [...] Or, [...] think about the archetypaltool you get in Minecraft. It’s a pickaxe. It’s not a tricorder. It’s not aBlack and Decker multi-tool. It’s a pickaxe, and through its very pickaxe-ness - low tech, implied manual labor, etc. - it tells part of the story ofthe world it exists in. [...] As soon as you decide what a game asset is,you’re implying the narrative that allows it to exist and function. [23]


One way to conceptually model feedback is per a typology (Table 3) importedfrom linguistics [20], which parallels Genette’s [39] tripartite model of narrative:the story, the discourse, and the narration.

Table 3. Typology of feedback available for narrative design.

Feedback Examples (“Feedback of this type includes...”)

Phonological Textual symbols, lines, shapes, haptics, sounds, lights, colors.Lexical Words, images, vibration patterns, voices, music notes, sound effects.Grammatical Texts, image sequences, camera shots, dialogue, music.Denotational Description, exposition, narration, characterization.

Phonological feedback is at the level of narration: the sounds, signs, and hap-tics that can be structured to convey meaning. Lexical feedback is at a higher-level of meaning within narration: a language inventory of the smallest unitsof meaning. In SMB, phonological colors contribute to Fig. 1’s depiction of theaxe, which is lexical. Grammatical feedback is at the level of discourse: stimulistructured according to a corresponding syntax. Adherence to that syntax facil-itates story sensemaking [18] and licenses inferences about underlying meaning.In SMB, image sequences are structured from left-to-right, which Grammaticallysuggests that the player progresses by going right-ward, potentially giving a clueon how to defeat Bowser for players who have never faced him before. Finally,denotational feedback is at the level of story: it includes stimuli that communi-cates the plot’s event structure, and it most-closely matches the discussed senseof “explicit narrative.” Fig. 2 depicts exposition that is denotational: it signalsto the player that their gameplay has concluded.

The content of Fig. 2’s feedback reinforces a particular interpretation of theplayer’s activity. This interpretation and the process that gives rise to it is thelast element of GFI, which we discuss next.

3.3 Interpretation

Interpretation is both: (a) the situated process of deriving meaning from en-action [76], and (b) the outcome of that process. When discussing interpretationin games, what is usually meant is the outcome of the player’s game experience.We typically want to answer: what is the narrative of a game?

Is it the space of potential narratives [54] the system affords? What of gamesabout story creation [56]? In describing the interpretations of a game, we wantto move away from false dichotomies like ludology v. narratology [32, 35] towarda more nuanced understanding of how these mutually inform and constrain eachother. For narrative design, we propose that it is important to shift the discussionfrom the outcome of a player’s interpretation, to the process that gives rise tothe outcome. This process is what crystallizes the defining function of narrativedesign: structuring feedback relative to mechanics and goals aimed at guiding a


Fig. 3. The parallel goal hierarchies of SMB [19]. The suggested readings are: ToFinish SMB means to “Save the Princess” (Toadstool); to Remove-Bowser means“Defeat Bowser”; and to Reach-Axe means “Destroy the Bridge with an Axe.”

player’s existing interpretation of the game’s narrative into a preferred interpre-tation. In this definition, the narrative designer (or whomever they represent) iswho prefers the interpretation. Further, “structuring” is broad: narrative designmay encompass changes to mechanics and goals, so long as these are intendedto effect changes to the interpretation that players may derive.

The interpretation outcome is what defines a narrative goal: the meaningderived from a ludological goal (§3.1) as experienced in a game. The feedback inFig. 2 scaffolds the player’s interpretation of their accomplishment: the player’squest – a prominent element in primarily-narrative games [46] – is over. Thus,one way to narratively make sense of To Finish is as “Save the Princess.”

Because narrative goals are interpretations of ludological ones, they bothentail the Parallel Goal Hierarchies (Fig. 3). The Ludological side reflectsthe sub-ordinate goals needed to satisfy super-ordinate goals; i.e. the “how.” Intandem, the Narrative side reflects the super-ordinate goals needed to motivatethe sub-ordinate goals; i.e. the “why.” The ludological side is best read top-down(you finish SMB by removing Bowser) whereas the narrative side is best readbottom-up (you defeat Bowser for the purpose of saving the Princess).

The hierarchies and their mapping is relative to the individual, and dependson both the game and its surrounding context. For instance, our interpretationof SMB ’s To Finish as “Save the princess” is plausible due to its (discussed)quest-like nature but also due to the game’s packaging, which asks: “Do youhave what it takes to save the Mushroom Princess?” (Fig. 4).

However, interpretations can be fragile. For instance, nothing indicates thatthe player is in fact “destroying the bridge with an axe:” the correspondinganimation is of such low framerate that a player may walk away with an alternateinterpretation (e.g. the bridge retracted). Further, is the object at the location toreach even an axe? Its feedback suggests that via a prototypical [57] appearance,


Fig. 4. SMB ’s packaging asks: “Do you have what it takes to save the MushroomPrincess?,” which supports interpreting SMB ’s ultimate goal as “Save the Princess.”

but the authors disagree: the third author sees it as a lever. In both cases it is thetask of narrative designers to match feedback for a ludological goal to narrow theplayers’ potential (existing) interpretation down towards the designer’s intended(preferred) interpretation (conversely, a game designer might need to craft aludological goal that aligns with the narrative designer’s intended feedback).

Dena [26] offers one way to conceptually model interpretation: the SequenceMethod [58], which is an established method for television (TV) series narra-tive design. This method models the player’s interpretation as a reader-responseprocess [51], driven by unanswered-questions, with an (eventual) outcome ofanswers, possibly inciting curiosity [26, p.43]:

[The Sequence Method] divides the experience into a series of questionsfor the audience [...] the overall question introduced at the beginningand answered near the end, and [...] multiple short-term questions tokeep driving the audience’s interest. “[when] answered, the [TV] seriesis forced to either introduce new central questions or end.” [58, op. cit.]

This method has been endorsed by several narrative designers within the gamesindustry. Bryant and Giglio [15], who have designed for both movies and games,argue that the method is useful to structure objectives for level design. Further,Bernstein [10], who has designed for movies, games, and television, argues thatthis method works better than the 3-Act Structure [34] because it is objective-driven (“What’s going to happen next?”) and fits well within gameplay loops.

4 GFI as a Lens

GFI groups three analytical, separate, causally-linked, and perspective-dependentlenses (Fig. 5). Through them, we straightforwardly unpack several thorny nar-rative design issues that are potentially challenging to analyze with MDA alone;these analyses lend support to the utility of GFI.


Fig. 5. Goals (like mechanics) are closer to the designer perspective. Interpretation (likeaesthetics) is closer to the player. Feedback (like dynamics) bridges between these.

4.1 Ludonarrative Dissonance and other Forms of Incoherence

(Game designer) Hocking coined the term ludonarrative dissonance in describingplaying Bioshock [37]: “[it suffers from] a powerful dissonance between what it isabout as a game, and what it is about as a story” [45, p. 256] The term has sincebeen widely adopted, critiqued [80], and reformulated [81, 8]. In view of GFI, itmanifests via a mismatch between the ludological and narrative goal hierarchies.

For example, games with the ultimate goal To Prolong often suffer from acertain amount of incoherence when they also include a “campaign” or “storymode” with an overarching conclusive narrative goal. The incoherence arisesbecause there is no clear narrative goal that makes sense of the ultimate one.In Destiny [5], once players achieve the narrative goal of “Defeat Atheon” (thefinal boss of the games’ raid), they return to the gameworld where nothinghas changed and the raid remains available to complete. Even once all of thegame’s most significant narrative goals have been achieved (e.g. side-quests andsecondary missions), the player should still continue To Prolong their play. Thelack of an “infinite narrative” (that aligns with To Prolong) poses a challengeif we want to see games without this fundamental incoherence, and perhapsdirectly motivates the use of procedural narrative generation [62].

There is also often dissonance/incoherence in games with multiple endings.Nier: Automata [68] has 26 different endings and (although many are optional)the player is required to successively complete the first 5 in order to witness all ofthe scripted narrative [52]. This creates dissonance: the game “indicate[s] to theplayer that an ending has been reached, only to enable continued play afterward,while coding the post-end portion of the game not as something extraneous,repeated, or additional, but an actual part of the game” [6].

Each time the player feels like they’ve finished, they learn that its narrativecounterpart has not been fully realized: “[the game] continually deprives theplayer of a sense of narrative closure” [52] with successive playthroughs requiringa re-interpretation of the games ultimate narrative goal as new elements areintroduced. For example, “reaching the B ending is a matter of following thesame core narrative events from the perspective of the android 9S instead of 2B,with only small narrative additions and gameplay alterations” [52].

4.2 Edge-cases of the Parallel Hierarchy

What if the Parallel Goal Hierarchies are imperfectly mapped? We consider twocases: there is a ludological goal with no evident narrative goal and vice-versa.


When a ludological goal has no evident narrative goal, the player has nointerpretation and thus no way to know the ludological goal exists. The only wayfor this ludological goal to be achieved is for the player to meet it by chance.There are at least two contexts in which this happens regularly.

The first is via cheat codes and the second is through meta-reward structuressuch as secret trophies or achievements [42]. From the player’s perspective, theseludological goals are met, but the player had no way of anticipating them fromanything communicated via the game. Consequently, the player has no way tomake sense of what happens other than appealing to conventions of the medium(“PS4 games have trophies” or “Konami games often have the Konami code”).

Conversely, when a narrative goal has no evident ludological one, a playerhas no way of achieving said goal in the game. This can lead to player frustra-tion, confusion, or disappointment (“The game asks that I do this thing, butit’s impossible!”). This might be the result of a mistake or flaw in the game’simplementation. Perhaps a player is told to activate a light switch, but it doesn’twork due to a software bug [60]. Sometimes, it might be the result of purposefuldesign. (Narrative designer) Ramanan describes a scene in Before I Forget [1], agame that “takes place in the soft pastel-colored home of Sunita, a woman withearly onset dementia” [83] in which players were reasonably confused:

[Players can’t find] the bathroom, and [...] every door [they] open turnsup in the same place, no matter which... [Players] were trying to see asystem and a logic when dementia doesn’t have any.”Ramanan as citedby Webber [83].

4.3 Localization, Remakes, and Sequels

Localization, when a game is modified in order to be sold in a new market [66],is different from translation because:

...localizing a video game may involve making technically or culturallymotivated changes that go beyond its textual structure, such as modify-ing the game code to accommodate the graphical discrepancies betweensource language and target languages [...] or even adjusting the game’smarketing strategies. [22]

Thus, localization often results in transcreation: departing from the originalsource to an extent such that the target is significantly different [66].

In view of GFI, localization requires transforming a game’s Narrative GoalHierarchy to preserve the relationships between narrative goals and their ludo-logical counterparts. When done poorly, the intended meaning of the player’s ac-tivity can become opaque. Czech’s [22] study of the Polish game market demon-strates how the narrative goal of “obtaining a killing spree” – Removing (defeat-ing) a certain number of opposing players while Preventing your own Removal(defeat) – can become disassociated from its ludological goal, due to poor local-ization: it reads as obtaining “a series of victims” or “a series of donations.”


Osu! Tatakae! Ouendan! [48, Ouendan] is a rhythm-action game first releasedin Japan, later localized by the same developer into Elite Beat Agents [49, EBA]for the North American (NA) market. The developer aimed to provide an expe-rience comparable to Ouendan for NA audiences [59]. Interestingly,

...the localization team was not afraid to modify many of the aspects ofOuendan that were not part of the core gameplay. The result is a gamewith new characters and stories, [and] a new soundtrack – [...] one of themost relevant components in a rhythm game. Nevertheless, the Japanesegame and the localized version feel strikingly similar, as the gameplayis virtually unchanged. [...] [EBA] retained the concept, mechanics andgeneral atmosphere of [Ouendan], but involved a complete overhaul ofboth the textual and audiovisual elements. [59]

GFI explains how Ouendan and EBA are ludologically the same but narra-tively different: localization effectively replaced the narrative goal hierarchy, butmapped it onto the same ludological one.

5 GFI as a Tool

In addition to its analytical traction, GFI also has generative traction to helpus design games. Fig. 6 illustrates a game-centered interaction framework [4]; itcharts how GFI fills in gaps in the MDA model that must be filled to account fornarrative design-related phenomena. In it, the (game) System contains the Lu-dological Goals that must be achieved for players to succeed at the game. Theseare presented to the player via Feedback that the Player observes and interprets.Intepretation yields the player’s mentalization of Narrative Goals that motivatewhich tasks they end up pursuing, which forms part of their Aesthetic experi-ence. Players attempt to carry out those tasks by articulating them through thegame’s afforded Mechanics, which result in run-time Dynamics that perform anupdate on the underlying game’s System.

Under this framework, we propose that (1) interpretation is what should beconsidered as the end result that guides narrative design refinement, and (2) goals(with corresponding mechanics) and feedback are what should be refined to effectchange in that interpretation. How to do so is beyond our scope; we briefly chartdesign challenges around player expectations [65] that GFI helps us grapple with.

Fig. 6. An interaction framework for games that identifies how GFI fills in conceptualgaps in MDA that must be filled to account for narrative design.


5.1 Setting Up and Satisfying Player Expectations

Games are communicative acts [17]. In them, players/audiences expect design-ers/authors to “cooperate,” as in dialogue [41]: speakers are tacitly expected byhearers to be as detailed as they need to be, truthful, on-topic, and clear [40].Authors rely on or flaunt these expectations for communicative effect. For exam-ple, a seemingly random element of a story may have its purpose revealed laterin the discourse (e.g. Chekhov’s Gun [72]) or may lead audiences astray (the RedHerring [82]). The Parallel Goal Hierarchies give us the language to describe howdesigners can also setup and manipulate player expectations. Succinctly, play-ers might be offered feedback to scaffold certain narrative goal interpretations,creating expectations about gameplay via the ludological goals they motivate,which in turn may be satisfied or subverted.

At the edge of a particular waterfall within The Elder Scrolls V: Skyrim [11,Skyrim], players are prompted with feedback that sets up a player’s expectation:an audio cue with associated textual overlay that reads “Bard’s Leap SummitDiscovered.” Given the prompt’s timing (expectation of relevance) and content(expectation of detail), one rational interpretation is “Jump off the waterfall,”with the implied (and dangerous) ludological goal Reach-Base of waterfall.

The interpretation emerges because in Skyrim that kind of overlay appearswhenever a player enters a significant place (e.g. a city). Only if the playeracts to satisfy the implied ludological goal (i.e. by jumping off the waterfall),does the player land safely in a pool, encounter a ghostly bard who describestheir unsuccessful dive, and get an in-game skill-boost. That the player cannotencounter the ghost by simply exploring the base of the waterfall suggests adesigner’s intentional structuring of the feedback to elicit the narrative goal.

This is related to narrative affordances [16], action opportunities that playersimagine will continue their unfolding story. In GFI, this is tantamount to elicitingand satisfying a player’s expectations by way of a narrative goal that motivatesa ludological one, the latter then recognized (possibly rewarded) by the game.

5.2 Subverting and Shifting Player Expectations

In contrast, the story of Spec Ops: The Line (SpecOps, [29]) diverges fromthe conventional hero story. It leads U.S. Army Captain Walker – the player-controlled protagonist – into taking ethically fraught actions [53]. As it unfolds,the player’s interpretation of the ludological goals is shifted: expectations ofthe medium [14] (“You’re a hero, so hurting enemies is the right thing to do”)and of the genre [3] (third-person shooter) are narratively questioned and thensubverted (“You’re a war criminal who has just harmed civilians”).

This is possible because actions in stories can be functionally polyvalent [31]:in SpecOps, “Soldier harms enemies” narratively functions [70, 69] as “heroism”in the player’s initial interpretation and is shifted to function as “villainy” viafeedback that reinforces an anti-hero [75] interpretation.

Similarly, Brenda Romero (nee Brathwaite), leveraged multiple interpreta-tions in her boardgame Train [13] via the use of purposefully ambiguous me-chanics and game elements. In Train, players are tasked with efficiently loading


and delivering boxcars with yellow meeples to a terminal station only to learn thename of the destination towards the end: Auschwitz. Many players ”realize” theintended meaning of the game’s narrative goal, and then subversively act againstthe game’s lugolodical goals: “Some of these players would derail the cars, whileothers would create virtual ‘Denmarks’ to give refuge to the tokens” [13].

6 Conclusions

GFI supports systematic design iteration: anticipating how changes to the game’sstructure will manifest in effects on players. By traversing GFI’s three levels ofabstraction, we expect designers can better conceptualize games, which mighthelp “control for undesired outcomes, and tune for desired behavior” [47].

Throughout this paper, we decomposed games writ large, including ones thatare not necessarily narratively-centered. This speaks to a fundamental claim thatGFI buys into: all games afford to be interpreted as stories. This is vacuouslytrue because we are narratively intelligent [43] and as Aarseth [2] has stated:people can narrativise anything. Tetris [67] is sufficiently representational [7] toafford discussing as symbolizing American life [64] or bodies in a grave [55].

However, the point of articulating GFI is to suggest that, while all games“tell” stories, some stories are more intended than others. (Game designer)Bateman [7] argues that everything that is representational – i.e. all feedback– contributes to the narrative. Thus, a game’s potential for narrativization isproportional to the degree it communicates non-abstract information. The storyin SMB is interpretatively simpler than that of Skyrim, but they are both sto-ries nonetheless by virtue of being communicated via non-abstract feedback (i.e.phonological, lexical, grammatical, and denotational information). Thus, to elicitintended stories instead of alternate player-narrativizations, designers shouldcenter on manipulating the game’s feedback. Our future work will explore how.

By formally understanding game design, we are better able to analytically de-scribe particular game experiences, systematically investigate and predict causaldeterminants of those experiences, and better articulate the relevance of theseresearch efforts to game design practice. GFI helps make sense of how gameexperiences are intrinsically narrative ones; we use it to reject the long-standing“antagonistic” relationship between story and gameplay as a false dichotomy.

GFI bridges narrative and game design and development, interactive digi-tal narrative studies, and game research. In this paper, we use it to articulatewhat the activity of narrative design is and to explain phenomena that emergefrom this design activity. We have presented GFI as a framework of modellablecomponents, and have also presented models for each component; Goals aremodeled with our Ultimate/Imperatives typology from prior work [84, 25], Feed-back is modeled with a typology imported from linguistics summarized in Ta-ble 3, and Interpretation is modeled via the question-answering focused SequenceMethod [58]. We expect GFI to clarify and strengthen the iterative processes ofdevelopers, scholars, and researchers alike, facilitating the decomposition, study,and design of a broad class of game-based narrative experiences.


References

1. 3-Fold Games: Before I Forget. 3-Fold Games (2017), Video game published forthe Microsoft Windows and macOS platforms

2. Aarseth, E.: A narrative theory of games. In: Proceedings of the 6th InternationalConference on the Foundations of Digital Games. pp. 129–133. ACM, New York,NY, USA (2012)

3. Abell, C.: Genre, Interpretation and Evaluation. In: Proceedings of the 136th Meet-ing of the Aristotelian Society. vol. CXV, pp. 25–40 (2015)

4. Abowd, G.D.: Formal aspects of human-computer interaction. Ph.D. thesis, Uni-versity of Oxford (1991)

5. Activision: Destiny. Bungie, Santa Monica, California (2014), Video game pub-lished for various platforms

6. Backe, H.J.: Consecutive endings and the aesthetic potential of cognitive disso-nance. In: Proceedings of the 2018 Conference of the Digital Games ResearchAssociation: Extended Abstracts (2018)

7. Bateman, C.: Fiction denial and the liberation of games [limited edition workingpaper], bolton: University of bolton (2013)

8. Bateman, C.: No-one plays alone. Transactions of the Digital Games ResearchAssociation 3(2), 5–36 (2017)

9. Bates, J., et al.: The role of emotion in believable agents. Communications of theACM 37(7), 122–125 (1994)

10. Bernstein, J.: Reimagining Story Structure: Moving Beyond ThreeActs in Narrative Design (2013), presented at the Game Develop-ers Conference. Available at: https://www.gdcvault.com/play/1019675/

Reimagining-Story-Structure-Moving-Beyond. Last accessed: 2020-07-2011. Bethesda Game Studios: The Elder Scrolls V: Skyrim. Bethesda Softworks (2011),

Video game published for the Microsoft Windows platform12. Bjork, S., Holopainen, J.: Patterns in Game Design. Charles River Media, Hing-

ham, MA, USA (2005)13. Brathwaite, B., Sharp, J.: The mechanic is the message: A post mortem in progress.

In: Schrier, K., Gibson, D. (eds.) Ethics and Game Design: Teaching Values throughPlay, pp. 311–329. Information Science Reference, Hershey, New York (2010)

14. Brice, M.: Death of the player (10 2013), http://www.mattiebrice.com/

death-of-the-player/ [Accessed: 04-04-19]15. Bryant, R.D., Giglio, K.: Slay the Dragon: Writing Great Video Games. Michael

Wiese Productions (2015)16. Cardona-Rivera, R.E.: A Model of Interactive Narrative Affordances. Ph.D. thesis,

North Carolina State University (2019)17. Cardona-Rivera, R.E., Young, R.M.: Games as Conversation. In: Proceedings of

the 3rd Workshop on Games and NLP at the 10th AAAI Conference on ArtificialIntelligence and Interactive Digital Entertainment. pp. 2–8 (2014)

18. Cardona-Rivera, R.E., Young, R.M.: Desiderata for a Computational Model ofHuman Online Narrative Sensemaking. In: the Working Notes of the 2019 AAAISpring Symposium on Story-enabled Intelligence (2019)

19. Cardona-Rivera, R.E., Zagal, J.P., Debus, M.S.: Narrative Goals in Games: ANovel Nexus of Story and Gameplay. In: Proceedings of the 15th InternationalConference on the Foundations of Digital Games (2020)

20. Cohn, N.: Your brain on comics: A cognitive model of visual narrative comprehen-sion. Topics in Cognitive Science (2019). https://doi.org/10.1111/tops.12421


21. Currie, G.: Arts and Minds. Oxford University Press (2004)22. Czech, D., et al.: Challenges in video game localization: An integrated perspective.

Explorations: A Journal of Language and Literature 1, 3–25 (2013)23. Dansky, R.: Screw narrative wrappers (2014), https://www.gamasutra.com/

blogs/RichardDansky/20140623/219615/Screw_Narrative_Wrappers.php [Ac-cessed: 02-03-20]

24. Debus, M.S.: Unifying Game Ontology: A Faceted Classification of Game Elements.Ph.D. thesis, IT University of Copenhagen, Denmark (2019)

25. Debus, M.S., Zagal, J.P., Cardona-Rivera, R.E.: A typology of imperative gamegoals. Game Studies 20(3) (2020, forthcoming)

26. Dena, C.: Finding a way: Techniques to avoid schema tension in narrative design.Transactions of the Digital Games Research Association 3(1) (2017)

27. Department, N.C.: Super Mario Bros. Nintendo, Tokyo, Japan (1985), Video gamepublished for the Nintendo Entertainment System

28. Despain, W.: Writing for video game genres: From FPS to RPG. CRC Press (2009)29. Development, Y.: Spec Ops: The Line. 2K Games, Novato, California (2005), Video

game published for various platforms30. Dille, F., Platten, J.Z.: The ultimate guide to video game writing and design. Lone

Eagle Publishing Company (2007)31. Dolezel, L.: Occidental Poetics: Tradition and Progress. University of Nebraska

Press, Lincoln, NE, USA (1990)32. Eskelinen, M.: The gaming situation. Game studies 1(1), 68 (2001)33. Fernandez-Vara, C.: Game spaces speak volumes: Indexical storytelling. In: Pro-

ceedings of the 2011 Digital Games Research Conference: Think Design Play. Di-GRA (2011)

34. Field, S.: Screenplay: The Foundations of Screenwriting. Revised Ed. Bantam Dell(2005)

35. Frasca, G.: Ludologists love stories, too: notes from a debate that never took place.In: DiGRA conference (2003)

36. Freytag, G.: Freytag’s Technique of the Drama, An Exposition of Dramatic Com-position and Art. Scott, Foresman and Company, Chicago, IL, USA (1896), Trans.by Elias J. MacEwan from the German original Die Technik des Dramas (1863)

37. Games, I.: Bioshock. 2K Games, Novato, California (2007), Video game publishedfor various platforms

38. Gaver, W.: Designing for homo ludens. I3 Magazine 12(June), 2–6 (2002)39. Genette, G.: Narrative Discourse: An Essay in Method. Cornell University Press

(1980)40. Grice, H.P.: Logic and conversation. In: Syntax and semantics 3: Speech Acts, pp.

41–58. Elsevier (1975)41. Grice, H.P.: Meaning. The Philosophical Review 66(3), 377–388 (1957)42. Hamari, J., Eranti, V.: Framework for designing and evaluating game achievements.

In: Proceedings of the 2011 Conference of the Digital Games Research Association(2011)

43. Herman, D.: Storytelling and the Sciences of Mind. MIT Press (2013)44. Heussner, T., Finley, T.K., Hepler, J.B., Lemay, A.: The Game Narrative Toolbox.

Routledge (2015)45. Hocking, C.: Ludonarrative dissonance in bioshock: The problem of what the game

is about. Well played 1, 255–260 (2009)46. Howard, J.: Quests: Design, Theory, and History in Games and Narratives. AK

Peters/CRC Press, Wellesley, MA, USA (2008)


47. Hunicke, R., LeBlanc, M., Zubek, R.: MDA: A Formal Approach to Game Designand Game Research. In: Proceedings of the Workshop on Challenges in Game AIat the 19th National Conference on Artificial Intelligence (2004)

48. iNiS: Osu! Tatakae! Ouendan. Nintendo (2005), Video game published for theNintendo DS platform

49. iNiS: Elite Beat Agents. Nintendo (2006), Video game published for the NintendoDS platform

50. Interactive, B.: DayZ. Bohemia Interactive, Prague, Czech Republic (2018), Videogame published for the Windows platform

51. Iser, W.: Interaction between text and reader. In: Suleiman, S.R., Crosman, I.(eds.) The Reader in the Text, pp. 106–119. Princeton University Press (1980)

52. Jacevic, M.: “This. Cannot. Continue.”–Ludoethical Tension in NieR: Automata.In: Philosophy of Computer Games Conference 2017 (2018)

53. Jørgensen, K.: The positive discomfort of spec ops: The line. Game studies 16(2)(2016)

54. Koenitz, H.: Towards a theoretical framework for interactive digital narrative. In:Proceedings of the 3rd International Conference on Interactive Digital Storytelling.pp. 176–185 (2010)

55. Koster, R.: A Theory of Fun for Game Design. O’Reilly Media, Inc., Sebastopol,CA, USA (2013)

56. Kreminski, M.: Creativity support for story construction play experiences. In: Pro-ceedings of the 15th AAAI Conference on Artificial Intelligence and InteractiveDigital Entertainment. pp. 210–212 (2019)

57. Lakoff, G.: Cognitive models and prototype theory. In: Margolis, E., Laurence, S.(eds.) Concepts: Core Readings, pp. 391–421. MIT Press, Cambridge, MA, USA(1999)

58. Landau, N.: The TV Showrunner’s Roadmap: 21 Navigational Tips for Screenwrit-ers to Create and Sustain a Hit TV Series. CRC Press (2013)

59. Lepre, O.: Divided by language, united by gameplay: An example of ludologicalgame localization. In: Mangiron, C., Orero, P., Minako, O. (eds.) Fun for all: Trans-lation and accessibility practices in video games, pp. 111–128. Peter Lang Bern,Bern, Switzerland (2014)

60. Lewis, C., Whitehead, J., Wardrip-Fruin, N.: What went wrong: a taxonomy ofvideo game bugs. In: Proceedings of the fifth international conference on the foun-dations of digital games. pp. 108–115 (2010)

61. Linderoth, J.: Beyond the digital divide: An ecological approach to gameplay.Transactions of the Digital Games Research Association 1(1) (2013)

62. Martens, C., Cardona-Rivera, R.E.: Procedural Narrative Generation (2017), pre-sented at the Game Developers Conference. Available at: https://www.gdcvault.com/play/1024143/Procedural-Narrative. Last accessed: 2020-07-20

63. Mitchell, A., McGee, K.: Reading again for the first time: A model of rereading ininteractive stories. In: Proceedings of the 4th International Conference on Interac-tive Digital Storytelling. pp. 202–213 (2012)

64. Murray, J.H.: Hamlet on the Holodeck: The Future of Narrative in Cyberspace.MIT Press, Cambridge, MA, USA (2017)

65. Murray, J.H.: Why paris needs hector and lancelot needs mordred: Roles and func-tions for dramatic compression in interactive narrative. In: Proceedings of the 11thInternational Conference on Interactive Digital Storytelling. pp. 13–24 (2011)

66. O’Hagan, M., Mangiron, C.: Game localization. John Benjamins Publishing Com-pany, Amsterdam/Philadelphia (2013)


67. Pajitnov, A., Pokhilko, V.: Tetris. Independent (1984), Video game published forthe Commodore 64 and the IBM PC

68. PlatinumGames: Nier: Automata. Square Enix (2017), Video game published forthe Microsoft Windows, PlayStaion4, and XBox One platforms

69. Prince, G.: A Dictionary of Narratology: Revised Edition. University of NebraskaPress (2003)

70. Propp, V.: Morphology of the Folktale. University of Texas Press, Austin, TX,USA (1968)

71. Punday, D.: Involvement, Interruption, and Inevitability: Melancholy as an Aes-thetic Principle in Game Narratives. SubStance 33(3), 80–107 (2004)

72. Rayfield, D.: Anton Chekhov: A Life. Northwestern University Press (2000)73. Reed, A.A.: Changeful Tales: Design-Driven Approaches Toward More Expressive

Storygames. Ph.D. thesis, University of California Santa Cruz, Santa Cruz, CA,USA (2017)

74. Roth, C., Koenitz, H.: Towards creating a body of evidence-based interactive dig-ital narrative design knowledge: approaches and challenges. In: Proceedings of the2nd International Workshop on Multimedia Alternate Realities at the 10th ACMMultimedia Systems Conference. pp. 19–24 (2017)

75. Shafer, D.M., Raney, A.A.: Exploring how we enjoy antihero narratives. Journalof Communication 62(6), 1028–1046 (2012)

76. Shapiro, L.: Embodied Cognition. Routledge, New York, NY, USA (2019)77. Skolnick, E.: Video game storytelling: What every developer needs to know about

narrative techniques. Watson-Guptill Publications (2014)78. Smith, H., et al.: Plans and Purposes How Videogame Goals Shape Player Be-

haviour. Ph.D. thesis, IT University of Copenhagen (2006)79. Stenros, J.: The game definition game: A review. Games and culture 12(6), 499–520

(2017)80. Summerley, R.K., et al.: Ludic Dysnarrativa: How Can Fictional Inconsistency In

Games Be Reduced? Ph.D. thesis, University of the Arts London and FalmouthUniversity (2017)

81. Tocci, J.: “You Are Dead. Continue?”: Conflicts and Complements in Game Rulesand Fiction. Eludamos. Journal for Computer Game Culture 2(2), 187–201 (2008)

82. Turco, L.: The Book of Literary Terms: The Genres of Fiction, Drama, Nonfiction,Literary criticism, and Scholarship. UPNE (1999)

83. Webber, J.E.: Why the best video games lie to you (2019), https://onezero.

medium.com/why-the-best-video-games-lie-to-you-67ae22f1412f [Accessed:02-03-20]

84. Zagal, J.P., Debus, M.S., Cardona-Rivera, R.E.: On the ultimate goals of games:Winning, finishing, and prolonging. In: Proceedings of the 13th International Phi-losophy of Computer Games Conference (2019)

rogelio e. cardona-rivera, ph.d. - gfi: a formal approach ......rogelio e. cardona-rivera1; 2, jos e...

Documents